Here are some helpful resources when scoping a project.
Requirements Definition is the first (and most important) step in any development effort. This is especially true when the project involves new technologies like Machine Learning, speech or vision. The Xively toolset gives structure to this stage of the IOT development process.
Xively – Getting Started with IOT (Adam Michelson, Fraser Macdonald)
Transforming a traditional non-IOT product into a connected IOT device is a nontrivial and risky exercise. Xively helps refine customer requirements, select tools/platform, and identify interoperability requirements.
Steps in the “journey” are:
- Conceptualization and Visualization (Model First approach, Product Launcher)
- Proof of Concept Hardware – point of confusion
- User Application
- CRM Integration (Salesforce)
- Support (of Things)
- Analytics Integration (entirely another topic)
- IOT Platform – allowing devices to safely and securely connect
Embedded Speech (recognition and synthesis) is a particularly sophisticated and challenging area for embedded development. A handful of vendors focus on this technology.
Sensory offers world class speech and vision technologies that can be embedded into mobile and other consumer electronic products. The technologies can be implemented across a variety of operating systems and DSPs.
Nervana Systems (Intel) https://en.wikipedia.org/wiki/Nervana_Systems
Movidius (Intel) https://www.movidius.com/
Apical (ARM) http://www.arm.com/products/graphics-and-multimedia
Research in the area of embedded speech:
Power efficient speech recognition (edge device) (Price, Glass, Chandrakasan)
Learning spoken language phonetic components (Lee, Glass)
Cloud data analysis / learning platforms
Large scale data infrastructure that spans on-premise and off-premise. Gives access to “unstructured” data. Consolidates data into “data lakes.”
Kafka (Jay Kreps)
Kafka is a data processing architecture focusing on stream (log) data. An example is Amazon Web Services “topic” publish/subscribe infrastructure. Kafka address the problem of large quantities of data originating from disparate sources which is needed by a variety of consumers.
Tableau Workbooks (data analytics examples)
Deep Learning Survey (Cloud-based ML)